基于对话结构的多轮对话生成模型

目前,多轮对话生成研究大多使用基于RNN或Transformer的编码器-解码器架构.但这些序列模型都未能很好地考虑到对话结构对于下一轮对话生成的影响.针对此问题,在传统的编码器-解码器模型的基础上,使用图神经网络结构对对话结构信息进行建模,从而有效地刻画对话的上下文中的关联逻辑.针对对话设计了基于文本相似度的关联结构、基于话轮转换的关联结构和基于说话人的关联结构,利用图神经网络进行建模,从而实现对话上下文内的信息传递及迭代.基于DailyDialog数据集的实验结果表明,与其他基线模型相比,该模型在多个指标上有一定的提升.这说明使用图神经网络建立的模型能够有效地刻画对话中的多种关联结构,从而有利于神经网络生成高质量的对话回复.     

Dialogue Systems as Proof Editors

This paper shows how a dialogue system for information-seekingdialogues can be implemented in a type-theory-based syntax editor,originally developed for editing mathematical proofs.The implementation gives a simple logical metatheory tosuch dialogue systems and also suggests new functions forthem, e.g., a local undo operation. The method developed provides alogically based declarative way of implementing simple dialoguesystems that is easy to port to new domains.     

对话管理中基于槽特征有限状态自动机的方法研究

对话系统的研究已经成为人机交互技术发展的新热点。而对话管理则是其中最重要的组成部分．该文在当前对话管理的各种实现方法的基础上,提出了一种基于槽特征的自动机设计方法,其中应用了状态压缩和状态集、动作集的子空间划分。并着重以确认过程为例,阐述了确认策略控制函数及其对对话过程的影响．文中还提出了一种树形的意图分层结构,并将这种分层结构应用于主题检测与主题切换,成功解决了多主题对话系统的主题切换问题．最后,实验表明该文提出的设计方案在策略控制、主题检测与主题切换等方面具有较好性能,同时也具有一定扩展性．     

文本情感对话系统研究综述

对话系统作为人机交互的重要方式,有着广泛的应用前景.现有的对话系统专注于解决语义一致性和内容丰富性等问题,对于提高人机交互以及产生人机共鸣方向的研究关注度不高.如何让生成的语句在具有语义相关性的基础上更自然地与用户交流是当前对话系统面临的主要问题之一.首先对对话系统进行了整体情况的概括.接着介绍了情感对话系统中的对话情...     

A Statistical Dialogue Analysis Model Based on Speech Acts for Dialogue Machine Translation

In some cases, to make a proper translation of an utterance in a dialogue, different pieces of contextual information are needed. Interpreting such utterances often requires dialogue analysis including speech acts and discourse analysis. In this paper, a statistical dialogue analysis model for Korean–English dialogue machine translation based on speech acts is proposed. The model uses syntactic patterns and n-grams of speech acts. The syntactic patterns include surface syntactic features which are related to the language-dependent expressions of speech acts. Speech-act n-grams are used to approximate the context of utterances. The key feature is the use of speech-act n-grams based on hierarchical recency. Experimental results with trigrams show that the proposed model achieves an accuracy of 66.87% for the top candidate and 82.35% for the top three candidates. It indicates that the proposed model based on hierarchical recency outperforms the model based on linear recency.     

Q2SM: 基于BERT的多领域任务型对话系统状态跟踪算法

基于管道的方法是目前任务型对话系统的主要构建方式,在工业界具有广泛应用,而对话状态跟踪(dialogue state tracking,DST)是任务型对话系统中的核心任务。面对传统的方法在多领域场景下表现较差的问题,该文结合语言模型预训练的最新研究成果,该文提出了一种基于BERT的对话状态跟踪算法Q2SM(query to state model)。该模型的上游使用了基于BERT的句子表征与相似度交互的槽判定模块,下游使用了一种面向对话状态跟踪任务的自定义RNN: DST-RNN。在WOZ 2.0和MultiWOZ 2.0两个数据集上的实验表明,Q2SM相比于之前的最好模型,分别在联合准确率和状态F₁值两个评价指标上提升了1.09%和2.38%。此外,模型消融实验验证了,DST-RNN相比于传统的RNN或LSTM,不仅可以提升评价指标值,还可以加快模型的收敛速度。     

基于BERT的强化语境与语义信息的对话幽默识别模型

幽默识别是自然语言处理的新兴研究领域之一。对话的特殊结构使得在对话中的幽默识别相较于短文本幽默识别更具有挑战性。在对话中,除了当前话语以外,上下文语境信息对于幽默的识别也至关重要。因此,该文在已有研究的基础上结合对话的结构特征,提出基于BERT的强化语境与语义信息的对话幽默识别模型。模型首先使用BERT对发言人信息和话语信息进行编码,其次分别使用句级别的BiLSTM、CNN和Attention机制强化语境信息,使用词级别的BiLSTM和Attention机制强化语义信息。实验结果表明,该文方法能有效提升机器识别对话中幽默的能力。     

限定领域口语对话系统中超出领域话语的对话行为识别

由于领域外话语具有内容短小、表达多样性、开放性及口语化等特点,限定领域口语对话系统中超出领域话语的对话行为识别是一个挑战。该文提出了一种结合外部无标签微博数据的随机森林对话行为识别方法。该文采用的微博数据无需根据应用领域特点专门收集和挑选,又与口语对话同样具有口语化和表达多样性的特点,其训练得到的词向量在超出领域话语出现超出词汇表字词时提供了有效的相似性扩展度量。随机森林模型具有较好的泛化能力,适合训练数据有限的分类任务。中文特定领域的口语对话语料库测试表明,该文提出的超出领域话语的对话行为识别方法取得了优于最大熵、卷积神经网络等短文本分类研究进展中的方法的效果。     

Knowledge-Combining Methodology for Dialogue Design in Spoken Language Systems

In this paper, we propose a strategy for designing dialogue managers in spoken dialogue systems for a restricted domain. This strategy combines several information sources intuition, observation and simulation, in order to maximize the adaptation within the system capability and the expectation of the user. These sources are combined by an iterative process consisting of five steps, where different dialogue alternatives are proposed and evaluated sequentially. The evaluation process includes different measures depending on the information required. Several measures are proposed and analyzed in each step. We also describe a user-modeling technique and an approach for designing the confirmation sub-dialogues based on recognition confidence measures. The knowledge-combining methodology is described and applied to a railway information system. In a subjective evaluation, users from the university gave the system a 3.9 score on a 5-point scale with an average call duration of 205 seconds. The employers of the railway company were more critical of the system. They gave it a score of 2.1 even though the system resolved more than half of the calls (57.8%) within an average call duration of three minutes (185 seconds).     

Spontaneous Speech Parsing in Travel Information Inquiring and Booking Systems

Grammar-based parsing is a prevalent method for natural language understanding(NLU)and has been introduced into dialogue systems for spoken language processing (SLP).A robust parsing scheme is proposed in this paper to overcome the notorious phenomena,such as garbage,ellipsis,word disordering,fragment ,and ill-form,which frequently occur in splien utterances,Keyword categories are used as terminal symbols,and the definition of grammar is extended by introducing three new rule types,by-passing,up-messing and overcrossing,in addition to the general rules called up-tying in this paper,and the use of semantic items simplifies the semantics extraction.The corresponding parser marionette,which is essentially a partial chart parser,is enhanced to parse the semantic grammar.The robust parsing scheme integrating the above methods has been adopted in an air traveling information service system called EasyFlight,and has achieved a high performance when used for parsing spontaneous speeches.     

陪护机器人口语对话系统设计与实现

针对陪护机器人口语对话系统的难点给出了相应的解决方案：以话题为单位的聊天机器人知识库构建规则、简单有效的句子相似度计算方法、准确高效的答案检索模块。通过对78个回合的实际聊天结果的统计分析,验证了解决方案的有效性;同时分析了方案的局限性。     

面向端到端的情感对话生成研究综述

人机对话作为人工智能的重要研究内容,受到了学术界和工业界的广泛关注.受到深度学习在自然语言处理成功应用的启发,越来越多的神经网络模型被研究者关注.其中基于端到端的神经网络模型能够从大规模语料中学习到有价值的规律和特征,生成有意义且多样性的回复,被广泛地应用于情感对话生成研究中.面向基于端到端模型的情感对话生成研究展开综...     

一种基于弱监督学习的论坛帖子对话行为分类方法

论坛帖子对话行为分类可以明确每个帖子在当前线索中的角色,有助于重构论坛线索中的对话关系,提高论坛信息检索的效果。该文提出了一种基于弱监督学习的论坛帖子对话行为分类方法,把帖子的对话行为分类作为线索的序列标注问题来解决。该方法的特点是只要指定合理的特征约束,就可以训练对话行为分类模型。方法在CNET和edX数据集上的分类精确率分别达到75.6%和60.7%,优于有监督的条件随机域方法。     

Incremental hierarchical construction of modular supervisors for discrete-event systems

Adoption of the supervisory control framework introduced by Ramadge and Wonham has been slowed somewhat by the problem of state space explosion that arises in systems of the scale common to most industrial applications. Hierarchical and modular approaches have been explored historically as means for addressing this problem. The limitations of these approaches include that a ‘consistent’ hierarchy is often difficult to achieve, and that modular supervisors often conflict with each other when acting in conjunction. This paper offers an approach that addresses some of these issues by incrementally building modular supervisors that are nonconflicting by construction. Abstractions are employed to make the procedure more computationally feasible. Proof is given showing the set of modular supervisors generated in this manner meet given specifications without blocking. Furthermore, examples are provided that demonstrate the reduction in complexity that this approach provides.     

基于深度学习的电影知识智能对话

针对电影知识对话响应中涉及大量的专有名词(如人名、地名等)且一条响应中可能涉及到多条知识的特点,提出了一种新的电影知识智能对话方法。首先,构建电影知识库和电影知识对话数据集;然后,编码器部分采用3-hops记忆神经网络,将历史对话编码成向量;最后,解码器部分将GRU和3-hops记忆神经网络相结合,用GRU的隐藏状态作为3-hops的动态查询向量,指导响应生成。结果显示该电影知识智能对话方法不仅解决了生成无意义响应的问题,又能保证生成响应的多样性。     

Task structures as a basis for modeling knowledge-based systems

Recently, there has been an increasing interest in improving the reliability and quality of AI systems. As a result, a number of approaches to knowledge-based systems modeling have been proposed. However, these approaches are limited in formally verifying the intended functionality and behavior of a knowledge-based system. In this article, we proposed a formal treatment to task structures to formally specify and verify knowledge-based systems modeled using these structures. The specification of a knowledge-based system modeled using task structures has two components: a model specification that describes static properties of the system, and a process specification that characterizes dynamic properties of the system. The static properties of a system are described by two models: a model about domain objects (domain model), and a model about the problem-solving states (state model). The dynamic properties of the system are characterized by (1) using the notion of state transition to explicitly describe what the functionality of a task is, and (2) specifying the sequence of tasks and interactions between tasks (i.e., behavior of a system) using task state expressions (TSE). The task structure extended with the proposed formalism not only provides a basis for detailed functional decomposition with procedure abstraction embedded in, but also facilitates the verification of the intended functionality and behavior of a knowledge-based system. © 1997 John Wiley & Sons, Inc.     

知识对话系统的零资源在线更新

知识对话系统旨在使用外部知识和对话上下文生成符合客观事实的回复。已有知识对话的研究较少关注知识对话系统的在线更新的问题。在知识对话系统在线更新中,面临因与知识配对的对话语料标注成本过高而导致零对话语料可用的问题。该文针对知识对话系统零资源更新问题,提出使用Pseudo Data进行模型的在线更新。首先,针对不同的场景,分析成因并提出了不同的Pseudo Data生成策略。 此外,该文在数据集KdConv上验证了当对话语料零资源时该文提出的方法的有效性。实验结果表明,使用Pseudo Data进行更新的模型在知识利用率、主题相关性上接近使用人类标注数据的在线更新模型,能有效使得知识对话系统在对话语料零资源的情况下完成在线更新。     

口语对话系统中对话管理方法研究综述

口语对话系统是人机交互领域的核心技术,也是实现和谐人机交互的重要途径,具有重大的研究意义和应用价值,其中的各项理论和技术的研究进展一直备受关注.较为全面深入地总结了对话管理及口语对话系统的研究进展和现状.首先阐述了口语对话系统中的主要研究问题,包括系统各模块的研究内容与关键技术、系统的可移植性和鲁棒性设计等;然后从理论模型、研究进展及可用性等角度系统地剖析了现有的多种口语对话管理策略;最后展望了未来可能的研究方向和亟待解决的问题.     

面向任务口语对话系统中不含槽信息话语的端到端对话控制

端到端(end-to-end)模型因其能有效避免传统管道式设计存在的错误传递与累积问题,成为了近年来口语对话系统(spoken dialogue system, SDS)的研究热点。在面向任务SDS的end-to-end对话控制中,处理携带任务领域语义信息(槽信息)的话语可以结合命名实体识别、数据库查询结果等语义特征,而不含槽信息的话语,由于缺乏领域语义信息以及表达多样,其有效对话控制仍然是一个挑战。该文提出一种融合“显式”话语特征和“隐式”上下文信息的end-to-end混合编码网络用于处理不含槽信息话语。具体地,在应用卷积神经网络(convolutional neural network, CNN)对“显式”话语序列提取得到的特征表达的基础上,通过构造和捕获对话序列中“隐式”的系统后台上下文信息,进一步丰富了系统动作分类模型的特征表达。在限定领域面向中文任务SDS中的评估结果表明,与传统的管道式SDS和经典的end-to-end SDS相比,该文的方案在不含槽信息话语的单回合处理以及对话段整体性能上都得到了显著提升。